Peter E. Holderrieth

I am a final-year PhD student at CSAIL at MIT, advised by Tommi Jaakkola working on deep generative models. During my PhD, I also interned at Meta AI, working with Yaron Lipman and Ricky Chen on the GenAI team, and at NVIDIA, working with Arash Vahdat and Julius Berner on the FastGen team.

Before MIT, I earned an MSc in Statistics and an MSc in Neuroscience at the University of Oxford supported by a Rhodes Scholarship where I worked with Yee Whye Teh and with Stephen Smith. I graduated with a BSc in Mathematics from the University of Bonn where I worked with Andreas Eberle on stochastic differential equations. In the past, I also worked or interned at BCG, at the Max Planck Institute, at several Biotech/AI startups (Cellarity, Genomics plc), and at the German Parliament. Originally, I am from the wonderful small town Schwaigern, Germany. Besides my work, I have a passion for writing music and playing guitar/piano in various jazz and rock bands.

Enabling practitioners

I also care deeply about making modern generative modeling accessible to researchers and practitioners:

Introduction to Flow Matching and Diffusion Models

I created and taught this MIT course on modern generative modeling. The lectures have received more than 100,000 YouTube views and have been used by researchers and practitioners around the world to learn about diffusion and flow-matching models.

Flow Matching Textbook

Together with Meta, I wrote an extensive guide to flow matching and developed an accompanying Python package for quickly implementing and experimenting with flow-matching models.